CrypFormBench is a new benchmark jointly covering symbolic and computational security to evaluate LLMs on five formal analysis capabilities, with results showing top model Claude-3.5 scores 48.7/100 and most models struggling on generation, transformation, and correction.
hub
Competition-Level Code Generation with AlphaCode
27 Pith papers cite this work. Polarity classification is still indexing.
27
Pith papers citing it
abstract
Programming is a powerful and ubiquitous problem-solving tool. Developing systems that can assist programmers or even generate programs independently could make programming more productive and accessible, yet so far incorporating innovations in AI has proven challenging. Recent large-scale language models have demonstrated an impressive ability to generate code, and are now able to complete simple programming tasks. However, these models still perform poorly when evaluated on more complex, unseen problems that require problem-solving skills beyond simply translating instructions into code. For example, competitive programming problems which require an understanding of algorithms and complex natural language remain extremely challenging. To address this gap, we introduce AlphaCode, a system for code generation that can create novel solutions to these problems that require deeper reasoning. In simulated evaluations on recent programming competitions on the Codeforces platform, AlphaCode achieved on average a ranking of top 54.3% in competitions with more than 5,000 participants. We found that three key components were critical to achieve good and reliable performance: (1) an extensive and clean competitive programming dataset for training and evaluation, (2) large and efficient-to-sample transformer-based architectures, and (3) large-scale model sampling to explore the search space, followed by filtering based on program behavior to a small set of submissions.
hub tools
citation-role summary
background 3
citation-polarity summary
roles
background 3representative citing papers
Introduces OPT* tasks and two training regimes (solver-guided online policy optimization with rank-based reward shaping and search-based offline RL) plus a theoretical link between search success and information extraction per budget unit, showing empirical gains in optimization-like reasoning.
Property-guided LLM program synthesis with counterexample feedback creates direct heuristics for PDDL planning domains that require far fewer generations and less evaluation cost than score-based baselines.
DuST self-trains LLMs for code generation by ranking their own test-time samples via sandbox execution and applying GRPO, improving judgment by +6.2 NDCG and single-sample pass@1 by +3.1 on LiveCodeBench.
MR-Coupler leverages functional coupling analysis and LLMs to generate valid metamorphic test cases for over 90% of tasks while detecting 44% of real bugs, outperforming baselines by 64.90% in validity and 36.56% in false-alarm reduction.
DBCooker automates synthesis of database native functions via LLM-guided characterization, coding plans, hybrid filling, and progressive validation, delivering 34.55% higher accuracy than baselines on SQLite, PostgreSQL, and DuckDB while generating functions absent from SQLite 3.50.
LLM evaluation for RTL generation identifies three performance tiers with frontier models reaching high synthesis quality and reveals systematic failure differences between proprietary and open models.
Massive activations are constant large values in LLMs that function as indispensable bias terms and concentrate attention probabilities on specific tokens.
Voyager achieves superior lifelong learning in Minecraft by combining an automatic exploration curriculum, a library of executable skills, and iterative LLM prompting with environment feedback, yielding 3.3x more unique items and 15.3x faster milestone unlocks than prior methods while generalizing技能
InCoder is the first generative model to directly perform zero-shot code infilling via bidirectional context from a masked-then-appended training scheme, matching left-to-right models on synthesis while improving on type inference, comment generation, and variable renaming.
ASTOR improves a single code LLM across four tasks by 9.0-9.5% over the best specialist and 7.5-12.8% over prior multi-task RL baselines via utility-driven data scheduling and adaptive KL regularization.
More capable LLMs and agents generate code with greater volume and architectural decay, following a Volume-Quality Inverse Law that neither functional correctness nor prompting mitigates.
Sparse replacement layers decompose the MLP and attention modules of a chess-playing transformer to reveal verifiable tactical reasoning pathways and parallel computation patterns.
PennyLang dataset of 3,347 PennyLane samples boosts LLM code generation success via RAG from 8.7% to 41.7% for Qwen 7B and 78.8% to 84.8% for LLaMa 4.
PRIME enables online process reward model updates in LLM RL using implicit rewards from rollouts and outcome labels, yielding 15.1% average gains on reasoning benchmarks and surpassing a stronger instruct model with 10% of the data.
MR-Adopt deduces input transformations from hard-coded MR test cases using LLMs, data-flow refinement, and output-relation selection to enable reuse with new source inputs.
DPOP is a new loss function that prevents DPO from lowering preferred response likelihoods and outperforms standard DPO on diverse datasets, MT-Bench, and enables Smaug-72B to exceed 80% on the Open LLM Leaderboard.
CodeT5+ is a flexible encoder-decoder LLM family for code pretrained with diverse objectives on multilingual corpora and initialized from existing LLMs, achieving state-of-the-art results on code generation, completion, math programming, and retrieval tasks including new SoTA on HumanEval with the 1
BLOOM is a 176B-parameter open-access multilingual language model trained on the ROOTS corpus that achieves competitive performance on benchmarks, with improved results after multitask prompted finetuning.
APE generates instruction candidates via LLM and selects the best by zero-shot performance of a second LLM, matching or beating human prompts on 19 of 24 NLP tasks.
Frontier AI models' no-CoT 50% task-completion time horizons have doubled yearly over six years, reaching over 3 minutes for GPT-5.5 with projections to 25 minutes by 2030.
AGPO adaptively sets trust-region size and exploration temperature from group reward dispersion, entropy, and KL drift, yielding higher scores than PPO and GRPO on nine math benchmarks under fixed token budget.
A-ProS uses a hybrid multi-model feedback framework with stateful refinement to improve success rates on competitive programming problems, achieving over 2x gains compared to baseline agent loops.
GenCluster scales test-time compute via large-scale generation, behavioral clustering, ranking, and round-robin submission to achieve IOI gold medal performance with the open-weight gpt-oss-120b model.
citing papers explorer
-
Step-by-Step Optimization-like Reasoning in LLMs over Expanding Search Spaces
Introduces OPT* tasks and two training regimes (solver-guided online policy optimization with rank-based reward shaping and search-based offline RL) plus a theoretical link between search success and information extraction per budget unit, showing empirical gains in optimization-like reasoning.
-
Property-Guided LLM Program Synthesis for Planning
Property-guided LLM program synthesis with counterexample feedback creates direct heuristics for PDDL planning domains that require far fewer generations and less evaluation cost than score-based baselines.
-
Voyager: An Open-Ended Embodied Agent with Large Language Models
Voyager achieves superior lifelong learning in Minecraft by combining an automatic exploration curriculum, a library of executable skills, and iterative LLM prompting with environment feedback, yielding 3.3x more unique items and 15.3x faster milestone unlocks than prior methods while generalizing技能
-
Think Fast: Estimating No-CoT Task-Completion Time Horizons of Frontier AI Models
Frontier AI models' no-CoT 50% task-completion time horizons have doubled yearly over six years, reaching over 3 minutes for GPT-5.5 with projections to 25 minutes by 2030.